CN114070676B - Method and device for reporting and receiving AI network model support capability and storage medium - Google Patents

Abstract

A method and a device for reporting and receiving AI network model supporting capability, a storage medium, user equipment and a base station are provided, wherein the method for reporting the AI network model supporting capability comprises the following steps: determining the capability of supporting the AI network model, wherein the capability of supporting the AI network model comprises whether channel estimation by adopting the AI network model is supported or not; and reporting the capability of supporting the AI network model by using uplink resources in the random access process, or triggering and reporting the capability of supporting the AI network model in a connected state. The technical scheme of the invention can enable the base station to acquire the relevant parameters of the user equipment side about the AI network model.

Description

AI network model support capability reporting, receiving method and device, storage medium

technical field

The present invention relates to the field of communication technologies, in particular to a method and device for reporting and receiving AI network model support capabilities, and a storage medium.

Background technique

In channel estimation, the purpose of applying the artificial intelligence (AI) algorithm is to compare the channel estimation process, that is, the process of estimating all channel values from the pilot, to the traditional image restoration/denoising process. Noisy deep learning algorithm to complete channel estimation.

Currently, channel estimation based on the AI network model is done on the user equipment side. The user equipment can know the performance of each AI network model configured and the size of the input and output.

However, how the base station learns the relevant parameters of the AI network model on the user equipment side is an urgent technical problem to be solved.

Contents of the invention

The technical problem solved by the present invention is how to enable the base station to know the relevant parameters of the AI network model on the user equipment side.

In order to solve the above technical problems, an embodiment of the present invention provides a method for reporting the support capability of the AI network model. The method for reporting the support capability of the AI network model includes: determining the capability to support the AI network model, and the capability to support the AI network model includes whether to support the use of The AI network model performs channel estimation; the uplink resource is used to report the capability of supporting the AI network model during the random access process, or the capability of supporting the AI network model is triggered to be reported in the connected state.

Optionally, the reporting the capability of supporting the AI network model by using the uplink resource in the random access process includes: reporting the capability of supporting the AI network model by using the time-frequency resource for sending the preamble.

Optionally, reporting the capability of supporting the AI network model by using the time-frequency resource for sending the preamble includes: determining a subset of ROs and its type, and the type of the subset of ROs includes The user equipment that does not support the AI network model initiates random access and is used for the user equipment that does not support the AI network model to initiate random access; determine the actually used RO subset according to the capability of the AI network model and the type of the RO subset , and use any RO in the subset of the actually used ROs to initiate random access.

Optionally, before determining the RO subset and its type, the determining further includes: receiving the RO subset and its type configured from the base station.

Optionally, reporting the ability to support the AI network model by using the time-frequency resource for sending the preamble includes: determining a preamble subset and its type, and the type of the preamble subset includes The user equipment that does not support the AI network model initiates random access and the user equipment that does not support the AI network model initiates random access; determine the actually used preamble subset according to the capability of the AI network model and the type of the preamble subset , and select a preamble in the actually used preamble subset to send.

Optionally, the reporting the capability of supporting the AI network model by using the uplink resource in the random access process includes: reporting the capability of supporting the AI network model by using message 3.

Optionally, the triggering to report the ability to support the AI network model in the connected state includes: receiving a support capability report trigger instruction from the base station, the support capability report trigger instruction is used to instruct to report the ability to support the AI network model ; In response to the support capability reporting trigger instruction, use the PDSCH scheduled by the PDCCH to report the capability of supporting the AI network model.

Optionally, the triggering to report the capability of supporting the AI network model in the connected state includes: detecting a support capability report trigger event, the support capability report trigger event including a partial bandwidth switching event; if the support capability report trigger event is detected event, use the PDSCH scheduled by the PDCCH on the part of the bandwidth after the handover to report the capability of supporting the AI network model.

Optionally, if the ability to support the AI network model indicates that it supports the use of the AI network model for channel estimation, receive an AI model size report instruction trigger instruction from the base station, and the support ability report trigger instruction is used to instruct to report all AI networks The input size of the model; in response to the AI model size report trigger instruction, the PDSCH scheduled by the PDCCH is used to report the input size of all AI network models.

Optionally, after reporting the ability to support the AI network model using uplink resources during the random access process, it further includes: if the ability to support the AI network model indicates that it supports using the AI network model for channel estimation, triggering the report Input dimensions for all AI network models.

In order to solve the above-mentioned technical problems, the embodiment of the present invention also discloses a method for receiving AI network model support capability. The method for receiving AI network model support capability includes: receiving the support AI network model reported by the user equipment using uplink resources during the random access process The capability of supporting the AI network model includes whether to support channel estimation using the AI network model; and configuring a demodulation reference signal for the user equipment according to the capability of supporting the AI network model.

Optionally, configuring the demodulation reference signal for the user equipment according to the capability of supporting the AI network model includes: configuring different densities for the user equipment supporting the AI network model and the user equipment not supporting the AI network model The demodulation reference signal, wherein the density of the demodulation reference signal configured for the user equipment supporting the adoption of the AI network model is lower than the density of the demodulation reference signal configured for the user equipment not supporting the adoption of the AI network model.

Optionally, before receiving the capability of supporting the AI network model reported by the user equipment using uplink resources during the random access process, the method further includes: sending the RO subset and its type configuration information to the user equipment, and the RO's The type of the subset includes being used for user equipment supporting the AI network model to initiate random access and for user equipment not supporting the AI network model to initiate random access.

Optionally, before receiving the ability to support the AI network model reported by the user equipment using uplink resources during the random access process, it further includes: sending the preamble subset and its type configuration information to the user equipment, the preamble The type of the subset includes being used for user equipment supporting the AI network model to initiate random access and for user equipment not supporting the AI network model to initiate random access.

Optionally, before receiving the ability to support the AI network model reported by the user equipment using uplink resources during the random access process, it further includes: sending a support capability report trigger instruction to the user equipment, and the support capability report trigger instruction uses Instructing the user equipment to report the capability of supporting the AI network model.

In order to solve the above technical problems, the embodiment of the present invention also discloses an AI network model support capability reporting device, the AI network model support capability reporting device includes: a capability determination module, used to determine the ability to support the AI network model, the support AI The capability of the network model includes whether to support channel estimation using the AI network model; the capability reporting module is used to report the capability of supporting the AI network model by using uplink resources in the random access process.

In order to solve the above-mentioned technical problems, the embodiment of the present invention also discloses an AI network model support capability receiving device. The AI network model support capability reception device includes: a capability reception module, which is used to receive the uplink resources used by the user equipment during the random access process. The reported ability to support the AI network model, the ability to support the AI network model includes whether to support the use of the AI network model for channel estimation; a configuration module, configured to configure demodulation for the user equipment according to the ability to support the AI network model reference signal.

The embodiment of the present invention also discloses a storage medium on which a computer program is stored. When the computer program is run by a processor, the steps of the method for reporting the AI network model support capability are executed, or the steps of the AI network model support capability are executed. The steps of the receive method.

The embodiment of the present invention also discloses a user equipment, including a memory and a processor, the memory stores a computer program that can run on the processor, and the processor executes the AI when running the computer program. Steps in the network model support capability reporting method.

The embodiment of the present invention also discloses a base station, including a memory and a processor, the memory stores a computer program that can run on the processor, and the processor executes the AI network when running the computer program. The steps of the model support capability receiving method.

Compared with the prior art, the technical solutions of the embodiments of the present invention have the following beneficial effects:

The technical solution of the present invention determines the ability to support the AI network model, the ability to support the AI network model includes whether to support the use of the AI network model for channel estimation; the ability to support the AI network model is reported using uplink resources during the random access process. In the technical solution of the present invention, the user equipment can report the ability to support the AI network model during the random access process, so that the base station can configure the demodulation reference signal for the user equipment according to the ability of the user equipment to support the AI network model, and realize optimal allocation of resources.

Further, a subset of ROs and its type are determined, and the types of the subset of ROs include random access for user equipment that supports the AI network model and random access for user equipment that does not support the AI network model. input; determine the actually used RO subset according to the capability of supporting the AI network model and the type of the RO subset, and use the actually used RO subset to initiate random access. In the technical solution of the present invention, the user equipment can indirectly indicate its ability to support the AI network model according to the type of the RO subset used by it to initiate random access, without occupying additional resources or signaling to report the AI network support model capabilities, saving resources and signaling overhead.

Further, use message 3 to report the capability of supporting the AI network model. In the technical solution of the present invention, the user equipment reports its ability to support the AI network model through message 3, and by multiplexing message 3, the ability to report the AI network model can be completed while initiating a random access request, saving resources and signaling overhead.

Description of drawings

1 is a flow chart of a method for reporting an AI network model support capability according to an embodiment of the present invention;

FIG. 2 is a schematic diagram of a specific application scenario according to an embodiment of the present invention;

FIG. 3 is a schematic diagram of another specific application scenario of the embodiment of the present invention;

FIG. 4 is a flow chart of a method for receiving an AI network model support capability according to an embodiment of the present invention;

5 is a schematic structural diagram of an AI network model support capability reporting device according to an embodiment of the present invention;

FIG. 6 is a schematic structural diagram of an AI network model support capability receiving device according to an embodiment of the present invention.

Implementation

As mentioned in the background art, how the base station obtains the relevant parameters of the AI network model on the user equipment side is a technical problem to be solved urgently.

The inventors of the present application conducted research on the prior art and found that if the user equipment (User Equipment, UE) supports AI-based channel estimation, the demand for demodulation reference signal density is not very large, that is, compared with the traditional channel estimation For the UE, the network side may configure a lower-density demodulation reference signal (Demodulation Reference Signal, DMRS) for the UE that supports AI-based channel estimation.

In the technical solution of the present invention, the user equipment can report the ability to support the AI network model during the random access process, so that the base station can configure the demodulation reference signal for the user equipment according to the ability of the user equipment to support the AI network model, and realize optimal allocation of resources.

Furthermore, in the technical solution of the present invention, the user equipment can indirectly indicate its ability to support the AI network model according to the type of the RO subset used by it to initiate random access, without occupying additional resources or signaling to report the The ability to support AI network models saves resources and signaling overhead.

In order to make the above objects, features and advantages of the present invention more comprehensible, specific embodiments of the present invention will be described in detail below in conjunction with the accompanying drawings.

FIG. 1 is a flowchart of a method for reporting an AI network model support capability according to an embodiment of the present invention.

The AI network model support capability reporting method shown in FIG. 1 can be used on the user equipment side. The method for reporting the AI network model support capability may specifically include the following steps:

Step S101: Determine the ability to support the AI network model, the ability to support the AI network model includes whether to support the use of the AI network model for channel estimation;

Step S102: Reporting the capability of supporting the AI network model by using uplink resources in the random access process, or triggering the reporting of the capability of supporting the AI network model in the connected state.

It should be noted that the sequence numbers of the steps in this embodiment do not represent a limitation on the execution order of the steps.

In this embodiment, the AI network model is used to perform channel estimation according to the input channel estimation matrix, which may be implemented by using any appropriate AI network model, for example, it may be a model obtained by training based on historical data.

In the specific implementation of step S101, if the user equipment is configured with an AI network model, it means that the user equipment has the ability to support the AI network model, that is, the user equipment supports the use of the AI network model for channel estimation. Conversely, if the user equipment is not configured with the AI network model, it means that the user equipment does not have the ability to support the AI network model.

In order for the base station to know the ability of the user equipment to support the AI network model, in the specific implementation of step S102, the user equipment may use uplink resources to report the ability to support the AI network model during the random access process, or trigger Report the ability to support the AI network model. That is to say, the base station actively reports during the random access process, and needs to be triggered to report again in the connected state. The specific triggering method can be signaling indication or event triggering.

In a specific implementation, the user equipment will send messages to the network side (Network) through the uplink resources configured by the base station during the random access process, such as message 1 (Msg1), message 3 (Msg3) and so on. Then, the user equipment can multiplex the uplink resource to report the capability of supporting the AI network model, so as to complete the reporting of the capability of the AI network model while completing the random access.

Referring to FIG. 2 together, during the random access process, the network side sends a system information block 1 (System Information Block, SIB), and SIB1 indicates resources for sending a preamble (Msg1). By reading the SIB1, the UE determines the resources used to send the preamble to the network side to indicate its intention to access the network. If the network side receives Msg1 correctly, it sends a random access response message (Msg2) scrambled with a random radio network temporary identity (Random Radio Network Temporary Identity, RA-RNTI) to the UE. After sending Msg1, the UE can use RA-RNTI to monitor Msg2 from the network to descramble the message. Msg2 may contain the resource indication that the UE sends Mgs3. Then, UE sends its identity and initial access establishment request (Msg3) to the network side through the uplink scheduling instruction in Msg2. Finally, the network side can notify the UE of the completion of the initial access process through Msg4.

That is to say, the UE can report the capability of supporting the AI network model through Msg1 or Mgs3.

In the embodiment of the present invention, the user equipment can report the ability to support the AI network model during the random access process, so that the base station can configure demodulation reference signals for the user equipment according to the ability of the user equipment to support the AI network model, and realize optimal configuration of resources.

In a non-limiting embodiment of the present invention, step S102 shown in FIG. 1 may specifically include the following steps: report the capability of supporting the AI network model by using the time-frequency resource for sending the preamble.

As mentioned above, the UE can send the preamble through the resources configured by the base station to send the preamble, and can also report the capability of supporting the AI network model, so that the base station can know the AI network model supported by the UE while receiving the preamble Ability.

Continuing to refer to FIG. 2 , after receiving the SIB1 from the network side, the UE knows the resources for sending the Msg1, and sends the Msg1 at the resource location where the Msg1 is sent and reports its ability to support the AI network model.

Further, the above-mentioned steps may specifically include the following steps:

Determine a subset of ROs and their types, where the types of the RO subsets include random access for user equipment that supports the AI network model and random access for user equipment that does not support the AI network model; according to The capability of supporting the AI network model and the type of the RO subset determine the actually used RO subset, and use the actually used RO subset to initiate random access.

In a specific implementation, the UE can acquire the related configuration of the Physical Random Access Channel Occasion (Physical Random Access Channel Occasion, RO) through the SIB1 message. The relevant configuration specifically includes the period size of the RO, the number of ROs in the time domain in each PRACH period, the number of ROs multiplexed in the frequency, and the like. RO refers to the time-frequency domain resource used to send the preamble.

In this embodiment, the UE can also determine the RO subset and its type through the SIB1 message. The subset of ROs and their types may be pre-configured by the base station. Different subsets of ROs occupy different frequency domain and/or time domain resources.

In a specific implementation, the UE may receive the RO subset and its type configured by the base station. The base station may carry the configured RO subset and its type in SIB1.

In this embodiment, the user equipment can indirectly indicate its ability to support the AI network model according to the type of the RO subset used by it to initiate random access, without occupying additional resources or signaling to report the support for the AI network model capability, saving resources and signaling overhead.

Referring to FIG. 3 together, the network side evenly divides the RO into two RO subsets (ie, RO subset 1 and RO subset 2) in frequency. In addition, RO subset 2 is configured to be used for UEs that support AI channel estimation to initiate random access, and RO subset 1 is configured for UEs that do not support AI channel estimation to initiate random access. If the UE selects the RO in RO subset 2 to initiate random access, that is, selects the RO in RO subset 2 to send Msg1, the network side considers that the UE supports AI channel estimation; otherwise, if the UE selects RO subset 1 If the RO in the UE initiates random access, the network side considers that the UE does not support AI channel estimation.

In an optional embodiment, the UE may also use a specific preamble (preamble) to report the capability of supporting the AI network model.

When the UE sends Msg1, it can select a preamble from 64 different preambles to send. In this embodiment, the base station may divide the preamble into different types of preamble subsets in advance, for example, two types of preamble subsets. One type of preamble subset is used by UEs that support AI estimation, and the other type of preamble subset is used by UEs that do not support AI estimation. In other words, the UE can indirectly inform the base station whether it supports the AI network model or not by using different types of preamble subsets.

It can be understood that the number of preambles in the preamble subset can be customized, which is not limited in this embodiment of the present invention.

Further, the preamble subset and its type may be pre-configured by the base station and sent to the UE.

In another non-limiting embodiment of the present invention, step S102 shown in FIG. 1 may specifically include the following steps: use message 3 to report the capability of supporting the AI network model.

Different from the foregoing embodiments, in the embodiment of the present invention, after receiving Msg2 from the network side, the UE knows the resources for sending Msg3, and sends Msg3 at the resource location where Msg3 is sent and reports its ability to support the AI network model .

In the embodiment of the present invention, the user equipment reports its ability to support the AI network model through message 3, and by multiplexing message 3, it can complete the reporting of the AI network model capability while initiating a random access request, saving resources and signaling overhead.

In a non-limiting embodiment, step S102 shown in FIG. 1 may include the following steps: receiving a support capability report trigger instruction from the base station, the support capability report trigger instruction is used to indicate the ability to report the AI network model support ; In response to the support capability reporting trigger instruction, use the PDSCH scheduled by the PDCCH to report the capability of supporting the AI network model.

Different from the previous embodiment in which the UE actively reports its ability to support the AI network model, in this embodiment, the base station sends a supporting capability reporting trigger instruction to the UE, and in response to the supporting capability reporting instruction, the UE uses the physical downlink control channel to The Physical Downlink Shared Channel (PDSCH) scheduled by the (Physical Downlink Control Channel, PDCCH) reports its ability to support the AI network model. That is to say, the UE in this embodiment reports its ability to support the AI network model only in response to an instruction from the base station.

In a specific implementation, the base station instructing the UE to report through the trigger instruction may refer to instructing the UE to report through the media access control (MediaAccess Control, MAC) control element (Control Element, CE) bearing the trigger instruction.

In an optional embodiment, the above steps may specifically include the following steps: detecting a support capability report trigger event, where the support capability report trigger event includes a partial bandwidth switch event; if the support capability report trigger event is detected, use the switch The latter part of the bandwidth reports the ability to support the AI network model.

Different from the UE reporting its ability to support the AI network model in response to a trigger instruction from the base station in the foregoing embodiments, the UE reports its ability to support the AI network model through event triggering in this embodiment. Specifically, the UE can detect a partial bandwidth (BandWidth Part, BWP) switching event. If the trigger event for reporting the support capability is detected, such as switching from BWP1 to BWP2, the UE uses the switched part of the bandwidth to report the AI network support model capabilities. Otherwise, the UE does not report its ability to support the AI network model.

In a specific example, the network side instructs the UE to switch the BWP through downlink control information (Downlink Control Information, DCI), that is, the DCI carries a specific bit (bit) to instruct the UE to switch the BWP. After the UE receives the BWP switching instruction, the UE switches the BWP, and reports the capability of supporting the AI network model through the PDSCH resource scheduled by the PDCCH on the new BWP.

By triggering an instruction to indicate a report or an event to trigger a report, the UE does not need to report when it does not have the ability to support the AI network model, saving resource overhead.

In yet another non-limiting embodiment of the present invention, the following steps may be included after step S102 shown in FIG. 1: if the capability of supporting the AI network model indicates that it supports the use of the AI network model for channel estimation, trigger reporting of all AI network models input size.

In this embodiment, after the UE reports to the base station that it has the ability to support the AI network model, the base station can instruct the UE to report the input size of the AI network model through a trigger instruction, and at the same time allocate uplink resources for the UE to report the input size of the AI network model. Then the UE can report the input size (size) of all AI network models on the uplink resources configured by the base station. In other words, since different AI network models have different input/output sizes, the UE may notify the network side of the input/output sizes of all AI network models.

That is to say, the UE can report the AI model input/output size to the network side through Msg1 or Msg3. That is, the UE can report the input/output sizes of all its AI network models to the network side through the PDSCH scheduled by the PDCCH.

Please refer to FIG. 4 , the embodiment of the present invention also discloses a method for receiving AI network model support capability. The method for receiving AI network model support capability can be used on the network side, such as a base station, that is, the network side can execute each of the methods. step.

The AI network model support capability receiving method may specifically include the following steps:

Step S401: Receive the capability of supporting the AI network model reported by the user equipment using uplink resources during the random access process, the capability of supporting the AI network model includes whether to support the use of the AI network model for channel estimation;

Step S402: Configure a demodulation reference signal for the user equipment according to the capability of supporting the AI network model.

In this embodiment, the base station may obtain the AI network model supporting capability of each user equipment during the random access process, so as to configure a demodulation reference signal (Demodulation Reference Signal, Demodulation Reference Signal, DMRS).

Further, step S402 shown in FIG. 4 may include the following steps: configuring demodulation reference signals with different densities for user equipment that supports the use of the AI network model and user equipment that does not support the use of the AI network model, wherein the support adopts The density of the demodulation reference signal configured for the user equipment of the AI network model is lower than the density of the demodulation reference signal configured for the user equipment that does not support the use of the AI network model.

In short, UEs that support AI network models can be configured with lower-density DMRS, thereby saving resource overhead on the network side.

In a specific embodiment, before step S201 shown in FIG. 2, the following steps may be further included: sending the RO subset and its type configuration information to the user equipment, and the RO subset type includes information for supporting AI The user equipment of the network model initiates random access and is used for the user equipment not supporting the AI network model to initiate random access.

In this embodiment, the base station may send the RO subset and its type to the UE in SIB1, and different types of RO subsets have different functions. The UE can indirectly indicate to the base station its ability to support the AI network model by sending preambles using different types of RO subsets. Correspondingly, the base station can determine the type of the RO subset used by the received preamble The preamble is the capability of the UE to support the AI network model.

In another specific embodiment, before step S201 shown in FIG. 2, the following steps may be further included: sending a support capability report trigger instruction to the user equipment, the support capability report trigger instruction is used to instruct the user equipment to report the The ability to support AI network models.

Different from the foregoing embodiments in which the base station determines the ability of the UE sending the preamble to support the AI network model according to the type of the RO subset used by the received preamble, in this embodiment the UE responds to the support capability report trigger The instruction reports its ability to support the AI network model, so the base station can directly obtain the UE's ability to support the AI network model. For example, the ability of the UE to support the AI network model is obtained directly through the bit value in Msg3.

Referring to FIG. 5 , the AI network model supporting capability reporting device 50 may include a capability determining module 501 and a capability reporting module 502 .

Among them, the capability determination module 501 is used to determine the capability to support the AI network model, and the capability to support the AI network model includes whether to support the use of the AI network model for channel estimation; the capability reporting module 502 is used to use uplink resources in the random access process Report the ability to support the AI network model.

Referring to FIG. 6 , the device 60 for receiving AI network model support capabilities may include a capability receiving module 601 and a configuration module 602 .

Wherein, the capability receiving module 601 is used to receive the capability of supporting the AI network model reported by the user equipment using the uplink resources during the random access process, and the capability of supporting the AI network model includes whether to support the use of the AI network model for channel estimation; the configuration module 602 is configured to configure a demodulation reference signal for the user equipment according to the capability of supporting the AI network model.

In the embodiment of the present invention, the user equipment can report the ability to support the AI network model during the random access process, so that the base station can configure demodulation reference signals for the user equipment according to the ability of the user equipment to support the AI network model, and realize optimal configuration of resources.

For more details about the working principles and working methods of the AI network model support capability reporting device 50 and the AI network model support capability receiving device 60 , please refer to the related descriptions in FIGS. 1 to 4 , which will not be repeated here.

The embodiment of the present invention also discloses a storage medium, the storage medium is a computer-readable storage medium, and a computer program is stored thereon, and the steps of the method shown in FIG. 1 or FIG. 4 can be executed when the computer program is running. The storage medium may include ROM, RAM, magnetic or optical disks, and the like. The storage medium may also include a non-volatile memory (non-volatile) or a non-transitory (non-transitory) memory, and the like.

The embodiment of the present invention also discloses a user equipment. The user equipment may include a memory and a processor, and a computer program that can run on the processor is stored in the memory. When the processor runs the computer program, it can execute the steps of the method shown in FIG. 1 . The user equipment includes, but is not limited to, terminal equipment such as mobile phones, computers, and tablet computers.

The embodiment of the present invention also discloses a base station. The base station may include a memory and a processor, and the memory stores computer programs that can run on the processor. The processor may execute the steps of the method shown in FIG. 4 when running the computer program.

Although the present invention is disclosed above, the present invention is not limited thereto. Any person skilled in the art can make various changes and modifications without departing from the spirit and scope of the present invention, so the protection scope of the present invention should be based on the scope defined in the claims.

Claims (17)

1. A method for reporting AI network model support capabilities, characterized in that, comprising: Determine the ability to support the AI network model, the ability to support the AI network model includes whether to support the use of the AI network model for channel estimation; Reporting the ability to support the AI network model by using uplink resources during the random access process; The ability to report the support AI network model by using uplink resources in the random access process includes: Reporting the capability of supporting the AI network model by using the time-frequency resource for sending the preamble; The reporting of the ability to support the AI network model by using the time-frequency resource for sending the preamble includes: Determine a subset of ROs and their types, where the types of the RO subsets include random access for user equipment that supports the AI network model and random access for user equipment that does not support the AI network model; according to The ability to support the AI network model and the type of the RO subset determine the actually used RO subset, and use any RO in the actually used RO subset to initiate random access; Or, determine the preamble subset and its type, the type of the preamble subset includes being used for user equipment supporting the AI network model to initiate random access and for user equipment not supporting the AI network model to initiate random access ; Determine the actually used preamble subset according to the capability of supporting the AI network model and the type of the preamble subset, and select the preamble in the actually used preamble subset to send. 2. The AI network model support capability reporting method according to claim 1, characterized in that, before determining the subset of RO and its type, it also includes: A subset of ROs and their types are received from the base station configuration. 3. The method for reporting the AI network model support capability according to claim 1, wherein the report of the ability to support the AI network model using uplink resources during the random access process comprises: Use message 3 to report the capability of supporting the AI network model. 4. The AI network model support capability reporting method according to claim 1, further comprising: Triggering the reporting of the ability to support the AI network model in the connected state includes the following steps: receiving a support capability report trigger instruction from the base station, where the support capability report trigger instruction is used to indicate the ability to report the support capability of the AI network model; In response to the support capability reporting trigger instruction, use the PDSCH scheduled by the PDCCH to report the capability of supporting the AI network model. 5. The AI network model support capability reporting method according to claim 1, further comprising: Triggering the reporting of the ability to support the AI network model in the connected state includes the following steps: Detecting a support capability report trigger event, where the support capability report trigger event includes a partial bandwidth switching event; If the trigger event of reporting the support capability is detected, the PDSCH scheduled by the PDCCH is used on the part of the bandwidth after switching to report the capability of supporting the AI network model. 6. The AI network model support capability reporting method according to claim 4 or 5, further comprising: If the capability of supporting the AI network model indicates that it supports the use of the AI network model for channel estimation, then report the capability of supporting the AI network model and the input sizes of all AI network models together. 7. The method for reporting the AI network model support capability according to claim 1, wherein after reporting the capability of supporting the AI network model using uplink resources in the random access process, the method further includes: If the ability to support the AI network model indicates that it supports the use of the AI network model for channel estimation, receive an AI model size report trigger instruction from the base station, and the AI model size report trigger instruction is used to instruct to report the input size of all AI network models ; In response to the AI model size reporting trigger instruction, use the PDSCH scheduled by the PDCCH to report the input sizes of all AI network models. 8. A method for receiving an AI network model support capability, comprising: In the process of random access, receiving the capability of supporting the AI network model reported by the user equipment using uplink resources, the capability of supporting the AI network model includes whether to support the use of the AI network model for channel estimation; Configuring a demodulation reference signal for the user equipment according to the capability of supporting the AI network model; The user equipment uses the time-frequency resource for sending the preamble to report the capability of supporting the AI network model; when the user equipment reports the capability of supporting the AI network model, it determines the subset of ROs and its type, and the RO The types of the subset include being used for user equipment that supports the AI network model to initiate random access and for user equipment that does not support the AI network model to initiate random access; according to the capability of supporting the AI network model and the RO The type of the subset determines the subset of ROs actually used, and uses any RO in the subset of ROs actually used to initiate random access; Or, determine the preamble subset and its type, the type of the preamble subset includes being used for user equipment supporting the AI network model to initiate random access and for user equipment not supporting the AI network model to initiate random access ; Determine the actually used preamble subset according to the capability of supporting the AI network model and the type of the preamble subset, and select the preamble in the actually used preamble subset to send. 9. The AI network model support capability receiving method according to claim 8, wherein the configuring the demodulation reference signal for the user equipment according to the capability of supporting the AI network model comprises: Configuring demodulation reference signals of different densities for user equipment that supports the use of the AI network model and user equipment that does not support the use of the AI network model, wherein the density of the demodulation reference signal configured for the user equipment that supports the use of the AI network model is low Based on the density of the demodulation reference signal configured for the user equipment that does not support the use of the AI network model. 10. The AI network model support capability receiving method according to claim 8, characterized in that before receiving the capability of supporting the AI network model reported by the user equipment using uplink resources during the random access process, the method further includes: Send the RO subset and its type configuration information to the user equipment. 11. The AI network model support capability receiving method according to claim 8, characterized in that before receiving the capability of supporting the AI network model reported by the user equipment using uplink resources during the random access process, the method further includes: Send the preamble subset and its type configuration information to the user equipment. 12. The AI network model support capability receiving method according to claim 8, characterized in that before receiving the capability of supporting the AI network model reported by the user equipment using uplink resources during the random access process, the method further includes: Sending a support capability report trigger instruction to the user equipment, where the support capability report trigger instruction is used to instruct the user equipment to report the capability of supporting the AI network model. 13. An AI network model support capability reporting device, characterized in that it comprises: A capability determination module, configured to determine the capability to support the AI network model, where the capability to support the AI network model includes whether to support the use of the AI network model for channel estimation; A capability reporting module, configured to use uplink resources to report the capability of supporting the AI network model during the random access process; The capability reporting module uses the time-frequency resource for sending the preamble to report the capability of supporting the AI network model; specifically, the capability reporting module determines the RO subset and its type, and the RO subset type includes Initiating random access for user equipment supporting the AI network model and for initiating random access for user equipment not supporting the AI network model; determine the actual A subset of the ROs used, and using any RO in the subset of the actually used ROs to initiate random access; Or, determine the preamble subset and its type, the type of the preamble subset includes being used for user equipment supporting the AI network model to initiate random access and for user equipment not supporting the AI network model to initiate random access ; Determine the actually used preamble subset according to the capability of supporting the AI network model and the type of the preamble subset, and select the preamble in the actually used preamble subset to send. 14. An AI network model support capability receiving device, characterized in that it comprises: A capability receiving module, configured to receive the capability of supporting the AI network model reported by the user equipment using uplink resources during the random access process, the capability of supporting the AI network model includes whether to support the use of the AI network model for channel estimation; A configuration module, configured to configure a demodulation reference signal for the user equipment according to the capability of supporting the AI network model; The user equipment uses the time-frequency resource for sending the preamble to report the capability of supporting the AI network model; when the user equipment reports the capability of supporting the AI network model, it determines the subset of ROs and its type, and the RO The types of the subset include being used for user equipment that supports the AI network model to initiate random access and for user equipment that does not support the AI network model to initiate random access; according to the capability of supporting the AI network model and the RO The type of the subset determines the subset of ROs actually used, and uses any RO in the subset of ROs actually used to initiate random access; Or, determine the preamble subset and its type, the type of the preamble subset includes being used for user equipment supporting the AI network model to initiate random access and for user equipment not supporting the AI network model to initiate random access ; Determine the actually used preamble subset according to the capability of supporting the AI network model and the type of the preamble subset, and select the preamble in the actually used preamble subset to send. 15. A storage medium on which is stored a computer program, characterized in that, when the computer program is run by a processor, the steps of the method for reporting the AI network model support capability described in any one of claims 1 to 7 are executed, or Executing the steps of the AI network model support capability receiving method described in any one of claims 8 to 12. 16. A user device, comprising a memory and a processor, the memory stores a computer program that can run on the processor, wherein the processor executes claims 1 to 1 when running the computer program. Steps in the method for reporting the AI network model support capability described in any one of 7. 17. A base station, comprising a memory and a processor, the memory stores a computer program that can run on the processor, wherein the processor executes claims 8 to 12 when running the computer program Steps in any one of the AI network model support capability receiving methods.

Images